MURAT YILDIRIM

MURAT YILDIRIMJul 1, 2026

I stopped treating Amazon PPC bid optimization as a manual task. Instead, I built a workflow around Claude. Here's how it works: • Export the Amazon Advertising bulk report. • Feed the file into Claude. • Set the target ACOS, minimum and maximum CPC, and a few guardrails. • Claude analyzes thousands of keywords and calculates bid recommendations using Revenue Per Click (RPC) logic. • Review the suggested changes. • Upload the optimized bulk file back to Amazon. The biggest advantage isn't that it saves time. It's consistency. Every keyword is evaluated using the same framework instead of making subjective decisions like "let's increase this bid by 10%." The workflow also protects branded keywords, applies safety limits to avoid aggressive bid swings, and even generates a rollback file so every change can be reversed if necessary. For large accounts, this turns hours of repetitive PPC work into a process that takes minutes. This is exactly how I think AI should be used in e-commerce—not to replace expertise, but to automate the repetitive analysis so you can spend your time making better business decisions. The next step is obvious: connecting Claude directly to the Amazon SP-API so bid optimization happens without exporting a single report. I break down this workflow step by step inside AI Seller Lab. Like the post. Comment "PPC" and I'll send you the details. #AmazonFBA #AmazonPPC #AmazonSeller #Ecommerce #AI #Claude #AmazonAds

MURAT YILDIRIMJun 27, 2026

I don't build Amazon listings anymore. I build systems that build them. One command. Hundreds of SKUs. Zero clicks inside Seller Central. Here's the pipeline. 1. Design Claude generates 300 transparent, print-ready t-shirt designs in a single batch. 2. Placement A Python script automatically places every design onto every mockup. Same position. Same scale. No manual adjustments. No drift. 3. Listing Copy Claude analyzes each design and writes: • SEO-friendly title • 5 bullet points • Product description Every listing is niche-aware, keyword-aware, and unique. 4. Image Hosting The finished mockups are uploaded automatically and converted into live image URLs ready for Amazon. 5. Amazon SP-API This is where most automation projects stop. The system creates: • Parent listing • Every child variation • Color relationships • Size relationships Then submits everything directly to Amazon through the SP-API. One design instantly becomes a complete variation family. No flat files. No Seller Central. No manual uploads. The entire workflow looks like this: → Read new designs → Generate mockups → Write listing copy with Claude → Upload images → Build parent & child variations → Publish through the Amazon SP-API What used to take an entire day now runs in minutes while I'm working on something that actually grows the business. The biggest shift wasn't learning the SP-API. It was changing how I think. Most sellers see a listing as something they have to create. I see a listing as the output of a system. Once the pipeline exists, adding the next 100 designs costs almost nothing. That's the leverage AI and automation create. I'm documenting the entire architecture, including: • SP-API authentication & listing submission • Parent/child variation structure • Automatic mockup placement with zero drift • Claude prompts for listing generation • The Python workflow that ties everything together Comment "AUTO" if you'd like the full breakdown. #AmazonFBA #AmazonSeller #SPAPI #PrintOnDemand #Automation #AI #Python #ClaudeAI

MURAT YILDIRIMJun 25, 2026

Most Amazon sellers spend months looking for their next product. What if an AI could screen hundreds of product ideas in a few hours? Recently, I came across a workflow that combines expired patents, Claude, and Alibaba manufacturers. And it's one of the more interesting product research systems I've seen. Here's how it works: Step 1: Find forgotten products. Search expired patents in the USPTO database. It's free. It's public. And it contains hundreds of thousands of products that were once commercially valuable. Focus on patents from small companies that: → Went bankrupt → Were acquired → Stopped renewing their patents These aren't failed products. They're products that lost the business behind them. Step 2: Let Claude do the research. Instead of manually reading patent documents, feed them into Claude in batches. Claude becomes your product analyst. It can: → Translate legal claims into plain English → Identify the customer problem being solved → Estimate manufacturing complexity → Flag regulatory risks → Check for existing competition → Evaluate market potential → Rank opportunities by likelihood of success The real advantage isn't finding patents. It's having Claude review hundreds of opportunities faster than a human team ever could. Step 3: Validate manufacturing. Take the patent drawings to Alibaba manufacturers. Most people overlook this part. Patent drawings often contain: → Dimensions → Materials → Components → Assembly details Enough information for a factory to understand what they're quoting. The workflow becomes: Expired Patent → Claude Analysis → Manufacturer Validation → Product Launch And the marketplaces are already there: → Amazon for demand-driven searches → Etsy for niche and personalized variations → Shopify for building a long-term brand Most people use AI to create more content. A smaller group is using AI to discover products hiding inside public databases. The interesting question isn't: "What should I sell next?" It's: "What opportunities can AI find that humans don't have time to look for?" Comment "PATENT" and I'll share the Claude workflow and prompts used for evaluating expired patents. #AmazonFBA #AmazonSeller #Shopify #Ecommerce #EtsySeller #ClaudeAI #ArtificialIntelligence #ProductResearch #PrivateLabel #Manufacturing #Alibaba #AIAgents #Entrepreneurship #BuildInPublic